Analysis of images in the discrimination of land cover, by processing radar satellite images

Abstract

In the evaluation of terrestrial coverage by means of remote sensors, we find a variety of techniques dedicated to be able to register the terrestrial coverage, one of them is the use of optical satellite images also known as multispectral images, these images are formed by different bands of spectral that depend on the optical sensor of the observation satellite, among the known bands we have: Panchromatic, Red, Green, Blue, Near Infrared, among others. The analysis of multispectral images is based on the grouping of these spectral bands, forming color images by grouping three spectral bands; One of the disadvantages presented by optical images in the acquisition is the presence of cloud cover that prevents the registration of land cover. Another of the techniques used is the use of the images provided by the Synthetic Aperture satellites, better known as radar images (SAR). These images have the advantage that they can record the land cover with the presence of cloud cover and also without the presence of sunlight, also presents the disadvantage that the image does not have the level of detail as the optical image, the radar image is a representation of the terrestrial coverage based on texture levels that differ according to the polarization in the acquisition, we have four types of polarization, the Horizontal Horizontal (HH), Horizontal Vertical (HV), Horizontal Vertical (VH) and Vertical Vertical (VV). The methodology proposed is characterized by generating a color image from a grouping of three radar images (three different polarizations) in order to improve the analysis of the radar image through a color interpretation, passing from having a grayscale image to a color image.